Methods, Compositions and Systems for Analyzing Imaging Data

Abstract

The present invention provides methods, compositions and systems for the analysis of imaging data, in particular, whole-animal imaging data acquired using microCT. Included in the invention are methods for registering and comparing test images to one or more reference images to identify and analyze anatomical features of interest. Also provided by the invention are methods and systems for efficient, semi-automatic and fully automatic methods for generating morphological statistics for anatomical features contained in imaging data. Libraries of images, including raw data acquired from imaging apparatuses as well as processed images, are also encompassed by the present invention.

1 . A method for comparing a query image of a test subject to a reference image of a reference subject, wherein said reference image is selected from a virtual histology library, and wherein said comparing comprises:
 (a) selecting an anatomical feature in said reference image, wherein said anatomical feature comprises landmark points;   (b) identifying corresponding landmark points in said query image; and   (c) registering said query image and said reference image using said landmark points, thereby comparing said query image to said reference image.   
     
     
         2 . The method of  claim 1 , wherein said comparing further comprises:
 (a) generating morphological statistics for a region comprising said landmark points in said reference image;   (b) generating morphological statistics for a region comprising said landmark points in said query image;   (c) calculating a similarity criterion for said morphological statistics for said reference image and said morphological statistics for said query image.   
     
     
         3 . The method of  claim 2 , wherein said similarity criterion is compared to a threshold value, and if said similarity criterion exceeds said threshold value, then said similarity criterion indicates that said region comprising said landmark points in said reference image correlates to said region comprising said landmark points in said query image. 
     
     
         4 . The method of  claim 3 , wherein said reference image is associated with a genotype, and wherein if said similarity criterion exceeds said threshold value, then said similarity criterion indicates that said test subject possesses said genotype, and wherein if said similarity criterion does not exceed said threshold value, then said similarity criterion indicates that said test subject does not possess said genotype. 
     
     
         5 . The method of  claim 3 , wherein said reference image is associated with a normal biological state, and wherein if said similarity criterion exceeds said threshold value, then said similarity criterion indicates that said test subject is in said normal biological state, and wherein if said similarity criterion does not exceed said threshold value, then said similarity criterion indicates that said test subject is not in said normal biological state. 
     
     
         6 . The method of  claim 3 , wherein said reference image is associated with a disease state, and wherein if said similarity criterion exceeds said threshold value, then said similarity criterion indicates that said test subject is in said disease state, and wherein if said similarity criterion does not exceed said threshold value, then said similarity criterion indicates that said test subject is not in said disease state. 
     
     
         7 . The method of  claim 6 , wherein said disease state comprises a developmental defect. 
     
     
         8 . The method of  claim 1 , wherein said test subject and said reference subject are selected from an ex vivo embryo, an ex vivo fetus, and a tissue sample. 
     
     
         9 . The method of  claim 8 , wherein said ex vivo embryo is a mouse embryo. 
     
     
         10 . A virtual histology library formed by compiling a plurality of reference images, wherein each of said reference images is produced by a method comprising:
 (a) obtaining a microCT image of a reference subject by a method comprising:
 i. incubating a sample from said reference subject in a first staining composition comprising a first staining agent, thereby producing a stained sample; 
 ii. suspending said stained sample in a liquid having a density lower than that of said stained sample; and 
 iii. scanning said stained sample in an X-ray computed tomography scanner to produce said microCT image of said stained sample; 
   (b) identifying landmark points in said microCT image;   (c) generating morphological statistics for a region around said landmark points; and   (d) processing said microCT image using said morphological statistics, thereby producing said reference image.   
     
     
         11 . A virtual histology library according to  claim 10 , wherein said generating said morphological statistics comprises applying a shape-based statistical model to said landmark points. 
     
     
         12 . A virtual histology library according to  claim 10 , wherein said landmark points identify a member selected from: forebrain, midbrain, hindbrain, heart, liver, neural tube, and lung. 
     
     
         13 . A virtual histology library according to  claim 12 , wherein said landmark points identify ventricle and atrial cavities of said heart. 
     
     
         14 . A virtual histology library according to  claim 10 , wherein said first staining agent is selected from osmium tetroxide and phosphotungstic acid. 
     
     
         15 . A method for indexing and retrieving stored images based on image content, said method comprising:
 (a) selecting a plurality of features from each of a plurality of reference images of at least one reference subject, said plurality of features corresponding to distinct anatomical features of said at least one reference subject;   (b) recording said plurality of features from said plurality of reference images;   (c) indexing said plurality features from said plurality of reference images, wherein said indexing is based on morphological statistics calculated for each of said plurality of features, and wherein said indexing forms a searchable library of said digital images;   (d) selecting a plurality of features from a query image;   (e) calculating morphological statistics for each of said plurality of features from said query image;   (f) searching said library using said morphological statistics for said query image; and   (g) retrieving at least one reference image from said library using a similarity criterion, wherein said similarity criterion is calculated from said morphological statistics from said reference image and said morphological statistics from said query image.   
     
     
         16 . The method of  claim 15 , wherein said plurality of reference images and said query image are microCT images. 
     
     
         17 . The method of  claim 15 , wherein said recording is accomplished using a computer implemented method. 
     
     
         18 . The method of  claim 15 , wherein said indexing comprises assigning each of said plurality of reference images to a group using said morphological statistics for said reference images. 
     
     
         19 . A computer implemented method for classifying a subject, said method comprising:
 (a) obtaining an image of said subject;   (b) selecting an anatomical feature of said image;   (c) determining a distribution of values for said anatomical feature;   (d) calculating test indices for each of said distribution of values in (c); and   (e) classifying said subject as normal or abnormal by comparing said test indices with reference indices stored in a virtual histology library, wherein said subject is classified as abnormal to an extent that there is a deviation of said test indices from said reference indices.   
     
     
         20 . The method of  claim 19 , wherein said subject is a mouse embryo.